NTARC Database¶
NTARC Database
Database
Daniel Ferreira, TU Wien
Why?
- Reading papers is time-exhaustive
- Each day more research, impossible to read everything
- No standardized methodologies (in network traffic research)
The NT Analysis Pipeline
Goals
- Easy to find unresearched topics
- Enables another perspective of the research
- Enables reproducibility
- Encourages standardization
NTARC Database
What Is It?
Collection of JSON files, include structured and unstructured data
We collect many parameters on what was previously researched
e.g.:
Who has used SVM for anomaly detection, and in which dataset?
Is the size of the packet often used to detect DDoS attacks?
What We Have
- 71 papers with NTARC.v1
- > 50 papers with NTARC.v2
- v1 and v2 have 33 papers in common
v1 is now deprecated, but not all content exists (yet) in v2
Documentation
Viewing the Database
How Can I Use It?
Which papers do anomaly detection?
with grep (v1)
with grep (v2)
with python (v2)
with python (v2)
or
with python (v2)
What Can It Be Used For?
- Which methods have already been applied?
- What datasets are being used?
- What are the main gaps in the field?
- Whatever else you can think of…
What Features Are Most Used?
Made with NTARC.v1
Reference: Ferreira, D. C., Vázquez, F. I., Vormayr, G., Bachl, M., & Zseby, T. (2017, August). A meta-analysis approach for feature selection in network traffic research. In Proceedings of the Reproducibility Workshop (pp. 17-20). ACM.
Why Should I Contribute?
By contributing, you will…
- increase long-time quality of Network Analysis research
- quickly learn what is important to read in papers
- become more critical of research
- write easier to comprehend papers
How Can I Contribute?
- Find a paper that deals with traffic analysis at network level (ask us if you need suggestions)
- Read the paper
- Curate the paper to an NTARC file
- Open a pull-request on Github to add to the database
Curating a Paper
Editor demo
Summary
- NTARC format
- Search for papers in the database
- Adding papers is good for everyone!